PNYX/reasoning_gym_lmeh

这是一个固定数据集的实现，用于reasoning-gym，并可在lm-evaluation-harness生态系统中使用。数据集旨在通过a-vert方法应用语义提取，每个任务生成100个样本，并去除了所有指示或提示。目前支持的任务有gsm_symbolic、polynomial_equations和complex_arithmetic。

This is an implementation of reasoning-gym into a fixed dataset to be used within the lm-evaluation-harness ecosystem. The dataset is intended to be used with semantic extraction via the a-vert method, with 100 samples generated per task, and all instructions or hints removed. Currently, the tasks supported include gsm_symbolic, polynomial_equations, and complex_arithmetic.